When comparing result-sets, identify columns by name (and not only by index)

[Seddryck]

Currently, when comparing result-sets, the first column of system-under-test is always compared to the first column of the assertion. This is relatively intuitive but is not always suitable. With the support of NoSQL database returning result-set in JSON, it's a huge limitation. Indeed, seralization doesn't guaratee the ordering of the members and missing members on the first row can screw up all your table.

NBi should support to define that the comparison should be done on the column'name and not on the column index.

[Seddryck]

• add support for result-set by name (and not only by index)
• refactor result-set comparison
• refactor single-row comparison
• extract method for value comparison to a new (or existing?) class
• add support in Xml (equalTo, resultSet, column, ...)

[Seddryck]

at the difference of comparison based on index, the comparison based on name will ignore all the columns that are not explicitly listed as keys or values.

To define the keys and values, you can use the equalTo element

<equalTo keys-names="Key" values-names="Value1, Value2">
...
</equalTo>

You can also use the column element

<equalTo>
   <column name="Key" role="key" type="text"/>
   <column name="Value" role="value" type="numeric"/>
</equalTo>

In a result-set you can assign a name to a column by using the new attribute column-name of the cell element. Only values specified in the first row ill be taken into account.

<resultSet>
  <row>
    <cell column-name="Value">2016-12-10</cell>
    <cell column-name="KeyField">10</cell>
  </row>
  <row>
    <cell>2016-02-02</cell>
    <cell>12</cell>
  </row>
</resultSet>

[Seddryck]

Preview available at https://ci.appveyor.com/project/Seddryck/nbi/build/1.15.0-ci.14

[Seddryck]

Available in beta: https://github.com/Seddryck/NBi/releases/tag/v1.15-beta